Couvrette v. Wisnovsky (Oregon 2025): Terminating Sanctions for 15 Fabricated AI Cases — Case Dismissed With Prejudice | Advocate Prakhar

Case at a Glance

Background

Joanne Couvrette and other plaintiffs brought suit involving a winery dispute. During the summary judgment phase, plaintiffs filed three separate briefs containing cross-motions and opposition materials. Defendants’ motion for sanctions identified a pattern of fabricated citations throughout all three briefs: citations to cases that simply did not exist, fabricated quotations attributed to real cases that contained no such language, and a fabricated quotation falsely attributed to the Restatement (Second) of Contracts.

The court’s analysis counted 15 entirely non-existent cases and fabricated language attributed to 8 legitimate authorities. The court deliberately declined to identify the fictitious cases by name in its opinion, stating that naming them could “inadvertently amplify the lie that these cases exist.” Both plaintiffs and their attorneys (Mr. Brigandi and Mr. Murphy) were found to bear responsibility for the misrepresentations.

The court found that plaintiffs and their counsel had not been adequately forthcoming, candid, or apologetic about their conduct throughout the sanctions proceedings, which influenced the decision to impose the most severe available sanctions.

The AI Issue

The court characterized this case as a “notorious outlier in both degree and volume” in the expanding universe of AI hallucination sanctions cases. The simultaneous appearance of 15 fabricated cases across three separate briefs, spanning multiple rounds of summary judgment briefing, pointed unmistakably to systematic AI hallucination without any verification workflow. The court emphasized that plaintiffs shared responsibility with their attorneys — a significant finding that suggests clients who provide AI-generated content to their lawyers, or who direct AI-assisted drafting, cannot isolate themselves from the resulting sanctions.

What the Court Decided

• Filing briefs containing 15 non-existent cases and fabricated quotations attributed to real authorities violates Rule 11(b)(2) and the court’s inherent authority to sanction conduct that undermines the integrity of proceedings.
• Both plaintiffs and their counsel bear responsibility for misrepresentations in court filings — clients cannot insulate themselves from sanctions by attributing AI-generated content solely to their attorneys.
• The lack of candor and contrition during sanctions proceedings supported the most severe available sanction: dismissal with prejudice.
• The three briefs containing fictitious citations were stricken from the record entirely.
• Lead attorney Mr. Brigandi was ordered to pay monetary sanctions; second attorney Mr. Murphy was required to show cause why he should not also face sanctions.

“In the quickly expanding universe of cases involving sanctions for the misuse of artificial intelligence, this case is a notorious outlier in both degree and volume… If there was ever an ‘appropriate case’ to grant terminating sanctions for the misuse of artificial intelligence, this is it.”

— Magistrate Judge Clarke, D. Oregon, December 12, 2025

The India Angle

Indian Law Equivalent

The scale of fabrication in Couvrette — 15 non-existent cases across three filings — would likely be treated as criminal contempt in Indian courts under Section 2(c) of the Contempt of Courts Act, 1971. Beyond contempt, fabricating legal authorities on such a scale would constitute a fraud on the court — a concept recognized in Indian jurisprudence through cases like A.V. Papayya Sastry v. Government of AP (2007) 4 SCC 221, where the Supreme Court held that a judgment obtained by fraud or misrepresentation is a nullity. A verdict obtained on the basis of fabricated citations would be void ab initio.

Bar Council Rules

The scale of the Couvrette violations would trigger the most serious provisions of the BCI Rules. Rule 22 (accuracy of facts in pleadings), Rule 9 (not misleading the court), and Rule 14 (maintaining dignity of the court) would all be engaged. More significantly, the Supreme Court’s inherent power under Article 142 of the Constitution allows it to take direct action against advocates who abuse the process on this scale, independent of Bar Council disciplinary proceedings.

Practical Advice for Indian Advocates

• When using AI for summary judgment or complex motion briefing, adopt a citation-by-citation verification protocol where each citation is independently confirmed in a primary database before the brief is finalized.
• Clients who actively participate in AI-assisted drafting (e.g., providing AI-generated research to their advocate) should understand that they too may face sanctions — the responsibility is shared, not solely the advocate’s.
• If a brief has already been filed with suspected AI citations, conduct an immediate audit before the hearing date and seek leave to file a corrected brief with full explanation to the court.

Quick Takeaways

• 15 fake cases across 3 briefs resulted in case dismissal with prejudice and monetary sanctions on counsel.
• Both clients and attorneys can be held responsible for AI-generated fabrications in court filings.
• Lack of contrition during sanctions proceedings makes terminating sanctions more likely.

Deep Dive: Shared Responsibility and the Scale Problem in AI Hallucination Cases

Couvrette v. Wisnovsky raises two issues that distinguish it from most AI hallucination sanctions cases: scale and shared responsibility. On scale, 15 fictitious cases across three briefs is not a slip — it is a systematic failure. For this to occur, the attorneys involved must have had essentially no verification workflow at all. Every case cited in the briefs should have been confirmed in a legal database; none of the 15 fabricated cases would have survived that check for even a moment. The volume of the failure makes it impossible to characterize as an isolated oversight.

The shared responsibility finding is equally significant. Courts in AI hallucination cases have sometimes focused narrowly on the attorney who signed the brief, treating the client as an innocent victim of counsel’s negligence. Couvrette rejects that framing where the record suggests client involvement in the preparation of AI-assisted materials. This creates an important warning for clients who use AI tools directly to conduct legal research and then provide that research to their attorneys: the verification duty does not transfer entirely to counsel, and sanctions can reach the client independently.

The court’s deliberate decision not to name the fictitious cases also deserves note. By refusing to identify the fabricated case names, the court prevented those names from entering the legal record in a way that might perpetuate the hallucination — someone finding the court’s opinion might otherwise search for the fake cases and find the opinion referencing them. This is a thoughtful approach to information hygiene in an era where AI tools are trained on court opinions and legal databases.

For Indian advocates managing complex commercial litigation with multiple rounds of submissions — similar to the summary judgment briefing in Couvrette — the case establishes that scale is not a defense. Across India’s High Courts and the Supreme Court, senior advocates regularly delegate research to junior colleagues or to AI tools. Couvrette makes clear that this delegation does not dilute the signing advocate’s responsibility for verification. The brief that reaches the court must reflect the advocate’s personal knowledge that its citations are accurate, whatever the underlying research process.